FYI, we noticed a -16.8% improvement of hackbench.time.user_time due to commit: commit bedc1d70ce363e2eddab9de946bf1c724fb3337d ("richacl: Compute maximum file masks from an acl") https://git.kernel.org/pub/scm/linux/kernel/git/agruen/linux-richacl.git richacl-wip in testcase: hackbench on test machine: 24 threads Westmere-EP with 16G memory with following parameters: nr_threads: 1600% mode: process ipc: socket Hackbench is both a benchmark and a stress test for the Linux kernel scheduler. Disclaimer: Results have been estimated based on internal Intel analysis and are provided for informational purposes only. Any difference in system hardware or software design or configuration may affect actual performance. Details are as below: --------------------------------------------------------------------------------------------------> To reproduce: git clone git://git.kernel.org/pub/scm/linux/kernel/git/wfg/lkp-tests.git cd lkp-tests bin/lkp install job.yaml # job file is attached in this email bin/lkp run job.yaml ========================================================================================= compiler/ipc/kconfig/mode/nr_threads/rootfs/tbox_group/testcase: gcc-6/socket/x86_64-rhel-7.2/process/1600%/debian-x86_64-2016-08-31.cgz/lkp-ws02/hackbench commit: 4e0a9d33c4 ("richacl: Permission check algorithm") bedc1d70ce ("richacl: Compute maximum file masks from an acl") 4e0a9d33c43d7838 bedc1d70ce363e2eddab9de946 ---------------- -------------------------- fail:runs %reproduction fail:runs | | | %stddev %change %stddev \ | \ 16725596 ± 3% -6.4% 15649069 ± 2% hackbench.time.involuntary_context_switches 13761 ± 0% -1.2% 13599 ± 0% hackbench.time.system_time 656.07 ± 1% -16.8% 546.15 ± 2% hackbench.time.user_time 166139 ± 0% -3.2% 160883 ± 0% interrupts.CAL:Function_call_interrupts 424.00 ± 0% -100.0% 0.00 ± -1% vmstat.memory.buff 229619 ± 1% -8.0% 211297 ± 7% vmstat.system.cs 32428 ± 0% -8.4% 29717 ± 1% vmstat.system.in 5962 ± 4% -11.3% 5289 ± 0% proc-vmstat.nr_active_file 5962 ± 4% -11.3% 5289 ± 0% proc-vmstat.nr_zone_active_file 835.25 ± 12% +55.5% 1298 ± 27% proc-vmstat.numa_pte_updates 32608 ± 0% -100.0% 0.00 ± -1% proc-vmstat.pgpgin 2875 ± 0% -12.4% 2518 ± 12% turbostat.Avg_MHz 1.14 ± 1% +19.0% 1.36 ± 10% turbostat.CPU%c1 0.86 ± 1% +37.3% 1.19 ± 5% turbostat.CPU%c3 0.94 ± 1% +23.9% 1.17 ± 8% turbostat.CPU%c6 23846 ± 4% -11.3% 21162 ± 0% meminfo.Active(file) 5694722 ± 15% -27.1% 4152320 ± 0% meminfo.DirectMap2M 277350 ± 7% -89.0% 30420 ± 5% meminfo.DirectMap4k 1499 ± 27% -29.6% 1056 ± 1% meminfo.Mlocked 1499 ± 27% -29.6% 1056 ± 1% meminfo.Unevictable 11934 ± 5% -11.5% 10561 ± 0% numa-meminfo.node0.Active(file) 32089 ± 23% +59.9% 51299 ± 22% numa-meminfo.node0.AnonHugePages 11901 ± 4% -11.0% 10586 ± 0% numa-meminfo.node1.Active(file) 791.25 ± 24% -36.9% 499.50 ± 13% numa-meminfo.node1.Mlocked 791.25 ± 24% -36.9% 499.50 ± 13% numa-meminfo.node1.Unevictable 30583174 ± 3% -22.6% 23672117 ± 6% cpuidle.C1-NHM.time 7133823 ± 2% +94.7% 13886928 ± 12% cpuidle.C1E-NHM.time 114014 ± 5% +88.2% 214599 ± 8% cpuidle.C1E-NHM.usage 1.569e+08 ± 1% +32.4% 2.078e+08 ± 6% cpuidle.C3-NHM.time 264229 ± 2% +29.5% 342065 ± 6% cpuidle.C3-NHM.usage 2.264e+08 ± 0% +31.0% 2.967e+08 ± 8% cpuidle.C6-NHM.time 235969 ± 0% +31.0% 309198 ± 8% cpuidle.C6-NHM.usage 2983 ± 5% -11.5% 2640 ± 0% numa-vmstat.node0.nr_active_file 2983 ± 5% -11.5% 2640 ± 0% numa-vmstat.node0.nr_zone_active_file 2975 ± 4% -11.0% 2647 ± 0% numa-vmstat.node1.nr_active_file 195.75 ± 25% -37.0% 123.25 ± 14% numa-vmstat.node1.nr_mlock 195.75 ± 25% -37.0% 123.25 ± 14% numa-vmstat.node1.nr_unevictable 2975 ± 4% -11.0% 2647 ± 0% numa-vmstat.node1.nr_zone_active_file 195.75 ± 25% -37.0% 123.25 ± 14% numa-vmstat.node1.nr_zone_unevictable 1021 ± 11% -80.7% 196.86 ± 51% sched_debug.cpu.clock.stddev 1021 ± 11% -80.7% 196.86 ± 51% sched_debug.cpu.clock_task.stddev 7313 ± 21% -18.4% 5966 ± 14% sched_debug.cpu.curr->pid.stddev 0.00 ± 11% -79.9% 0.00 ± 48% sched_debug.cpu.next_balance.stddev 2191 ± 12% -36.0% 1402 ± 6% sched_debug.cpu.nr_load_updates.stddev 61.61 ± 25% +628.6% 448.91 ± 58% sched_debug.cpu.nr_uninterruptible.max -150.62 ±-39% +656.3% -1139 ±-86% sched_debug.cpu.nr_uninterruptible.min 42.65 ± 30% +871.0% 414.16 ± 76% sched_debug.cpu.nr_uninterruptible.stddev 0.93 ± 0% -12.9% 0.81 ± 0% perf-stat.branch-miss-rate% 4.047e+10 ± 0% -21.2% 3.19e+10 ± 10% perf-stat.branch-misses 1.633e+11 ± 0% -11.5% 1.445e+11 ± 9% perf-stat.cache-references 1.443e+08 ± 1% -8.8% 1.316e+08 ± 8% perf-stat.context-switches 4.28e+13 ± 0% -13.2% 3.717e+13 ± 13% perf-stat.cpu-cycles 2.05 ± 3% +7.1% 2.20 ± 5% perf-stat.dTLB-load-miss-rate% 0.19 ± 0% +9.4% 0.21 ± 1% perf-stat.dTLB-store-miss-rate% 0.00 ± 3% -18.9% 0.00 ± 1% perf-stat.iTLB-load-miss-rate% 8.641e+08 ± 3% -26.5% 6.35e+08 ± 10% perf-stat.iTLB-load-misses 28480 ± 3% +23.0% 35029 ± 1% perf-stat.instructions-per-iTLB-miss 0.57 ± 0% +4.6% 0.60 ± 2% perf-stat.ipc 3.041e+10 ± 0% -10.8% 2.714e+10 ± 10% perf-stat.node-stores 1.81 ± 1% -16.6% 1.51 ± 3% perf-profile.calltrace.cycles-pp.copy_from_iter.skb_copy_datagram_from_iter.unix_stream_sendmsg.sock_sendmsg.sock_write_iter 6.43 ± 1% -10.4% 5.77 ± 4% perf-profile.calltrace.cycles-pp.rw_verify_area.vfs_read.sys_read.entry_SYSCALL_64_fastpath 5.58 ± 2% -9.1% 5.08 ± 5% perf-profile.calltrace.cycles-pp.security_file_permission.rw_verify_area.vfs_read.sys_read.entry_SYSCALL_64_fastpath 1.34 ± 1% +12.5% 1.51 ± 7% perf-profile.calltrace.cycles-pp.security_socket_sendmsg.sock_sendmsg.sock_write_iter.__vfs_write.vfs_write 0.93 ± 4% +13.4% 1.06 ± 10% perf-profile.calltrace.cycles-pp.selinux_socket_sendmsg.security_socket_sendmsg.sock_sendmsg.sock_write_iter.__vfs_write 4.97 ± 0% -17.6% 4.10 ± 3% perf-profile.calltrace.cycles-pp.skb_copy_datagram_from_iter.unix_stream_sendmsg.sock_sendmsg.sock_write_iter.__vfs_write 1.51 ± 0% +29.7% 1.96 ± 20% perf-profile.calltrace.cycles-pp.sock_def_readable.unix_stream_sendmsg.sock_sendmsg.sock_write_iter.__vfs_write 0.88 ± 4% +14.5% 1.00 ± 10% perf-profile.calltrace.cycles-pp.sock_has_perm.selinux_socket_sendmsg.security_socket_sendmsg.sock_sendmsg.sock_write_iter 1.78 ± 1% -27.2% 1.29 ± 4% perf-profile.children.cycles-pp.__might_fault 1.50 ± 1% +48.7% 2.23 ± 42% perf-profile.children.cycles-pp._raw_spin_lock_irqsave 1.09 ± 1% -19.5% 0.88 ± 1% perf-profile.children.cycles-pp._raw_spin_unlock_irqrestore 1.25 ± 3% -7.4% 1.16 ± 3% perf-profile.children.cycles-pp.apic_timer_interrupt 1.89 ± 1% -16.0% 1.58 ± 2% perf-profile.children.cycles-pp.copy_from_iter 1.11 ± 2% -10.4% 0.99 ± 5% perf-profile.children.cycles-pp.hrtimer_interrupt 1.13 ± 3% -10.0% 1.01 ± 5% perf-profile.children.cycles-pp.local_apic_timer_interrupt 1.40 ± 1% +12.2% 1.56 ± 6% perf-profile.children.cycles-pp.security_socket_sendmsg 1.00 ± 3% +12.7% 1.13 ± 9% perf-profile.children.cycles-pp.selinux_socket_sendmsg 5.07 ± 0% -16.8% 4.22 ± 3% perf-profile.children.cycles-pp.skb_copy_datagram_from_iter 1.57 ± 0% +28.1% 2.01 ± 20% perf-profile.children.cycles-pp.sock_def_readable 1.23 ± 1% -20.4% 0.98 ± 8% perf-profile.self.cycles-pp.__vfs_write 1.09 ± 1% -19.5% 0.88 ± 1% perf-profile.self.cycles-pp._raw_spin_unlock_irqrestore 1.26 ± 1% -14.1% 1.08 ± 2% perf-profile.self.cycles-pp.copy_from_iter 1.15 ± 0% -14.4% 0.98 ± 7% perf-profile.self.cycles-pp.kfree 2.07 ± 3% -24.5% 1.56 ± 5% perf-profile.self.cycles-pp.security_file_permission 1.06 ± 2% -11.3% 0.94 ± 2% perf-profile.self.cycles-pp.sock_wfree 682.25 ± 10% -94.3% 39.00 ± 0% slabinfo.bdev_cache.active_objs 682.25 ± 10% -94.3% 39.00 ± 0% slabinfo.bdev_cache.num_objs 1115 ± 1% -100.0% 0.00 ± -1% slabinfo.blkdev_requests.active_objs 1115 ± 1% -100.0% 0.00 ± -1% slabinfo.blkdev_requests.num_objs 433.00 ± 11% -100.0% 0.00 ± -1% slabinfo.buffer_head.active_objs 433.00 ± 11% -100.0% 0.00 ± -1% slabinfo.buffer_head.num_objs 229.50 ± 19% -100.0% 0.00 ± -1% slabinfo.ext4_extent_status.active_objs 229.50 ± 19% -100.0% 0.00 ± -1% slabinfo.ext4_extent_status.num_objs 1204 ± 0% -100.0% 0.00 ± -1% slabinfo.ext4_groupinfo_4k.active_objs 1204 ± 0% -100.0% 0.00 ± -1% slabinfo.ext4_groupinfo_4k.num_objs 703.75 ± 6% -72.6% 193.00 ± 11% slabinfo.file_lock_cache.active_objs 703.75 ± 6% -72.6% 193.00 ± 11% slabinfo.file_lock_cache.num_objs 4306 ± 0% -15.1% 3655 ± 0% slabinfo.ftrace_event_field.active_objs 4306 ± 0% -15.1% 3655 ± 0% slabinfo.ftrace_event_field.num_objs 256.00 ± 0% -100.0% 0.00 ± -1% slabinfo.jbd2_revoke_table_s.active_objs 256.00 ± 0% -100.0% 0.00 ± -1% slabinfo.jbd2_revoke_table_s.num_objs 39639 ± 0% -29.3% 28035 ± 0% slabinfo.kernfs_node_cache.active_objs 1165 ± 0% -29.3% 823.75 ± 0% slabinfo.kernfs_node_cache.active_slabs 39639 ± 0% -29.3% 28035 ± 0% slabinfo.kernfs_node_cache.num_objs 1165 ± 0% -29.3% 823.75 ± 0% slabinfo.kernfs_node_cache.num_slabs 3256 ± 1% -49.1% 1656 ± 1% slabinfo.kmalloc-128.active_objs 3256 ± 1% -49.1% 1656 ± 1% slabinfo.kmalloc-128.num_objs 4385 ± 4% -37.2% 2754 ± 0% slabinfo.kmalloc-192.active_objs 4406 ± 3% -37.1% 2772 ± 1% slabinfo.kmalloc-192.num_objs 1036 ± 3% -37.2% 651.00 ± 9% slabinfo.mnt_cache.active_objs 1036 ± 3% -37.2% 651.00 ± 9% slabinfo.mnt_cache.num_objs 1103 ± 6% -43.7% 620.50 ± 13% slabinfo.nsproxy.active_objs 1103 ± 6% -43.7% 620.50 ± 13% slabinfo.nsproxy.num_objs 2449 ± 2% -28.5% 1750 ± 3% slabinfo.shmem_inode_cache.active_objs 2449 ± 2% -28.5% 1750 ± 3% slabinfo.shmem_inode_cache.num_objs 2806 ± 1% -24.7% 2114 ± 2% slabinfo.trace_event_file.active_objs 2806 ± 1% -24.7% 2114 ± 2% slabinfo.trace_event_file.num_objs hackbench.time.user_time 680 ++--------------------------------------------------------------------+ | .*. | 660 ++ *.*. .*. .* * | 640 ++ .*. .*. .*. .. * *. | *.*.*. *.*.*.*.. .*.*.* *..*.*.* * | 620 ++ * | 600 ++ | | | 580 ++ O O | 560 O+O O O O O O O O O O O O | | O O O O O O O O O O 540 ++ O O | 520 ++ O O O O O O | | | 500 ++--------------------------------------------------------------------+ [*] bisect-good sample [O] bisect-bad sample Thanks, Xiaolong